Data centre connectivity: Do it yourself or outsource?

Only a handful of companies in the world can truly say that, when it comes to data centre connectivity, they do it all themselves. That is to say, if doing it yourself means owning and operating the complete, end-to-end infrastructure required to link up data centres located across the world, then virtually no organisation can do this.

No, the reality is that all manner of networks, owned by a variety of different organisations (telcos, service providers, data centre owners and the like) interlink to provide the connectivity required by any specific organisation.

The DIY Approach

More realistically, do it yourself data centre connectivity means providing the data centre infrastructure and the connectivity resiliency required by any particular organisation. So, 99.999% availability needs to be provided by large scale redundancy – redundant and separate power supplies per device, redundant cooling, and at least two separate network paths within a data centre – connected to the outside world via at least two separate locations in the data centre, and with these outside connections provided by different connectivity providers.

The cost of putting in place such an infrastructure – and one that meets the varying bandwidth and latency requirements of the different applications that are running on it – will be enough to deter most organisations from pursuing this approach. Especially when one considers the ongoing costs of equipment maintenance and renewal, not to mention the requirement for a level of in-house expertise that does not come cheap.

The advantages of the do it yourself approach? Difficult to find many – the more so as even government and financial organisations, who might traditionally have ‘gone it alone’ for security and/or strategic reasons, are beginning to embrace the world of colos, Cloud and managed services.

I suppose one might say that the do it yourself approach provides the impression of being in control. However, when one considers that the networks beyond the data centre walls are almost certainly owned by other organisations, I’m not sure how convincing is this sense of security.

The Outsourcing Approach

In reality, outsourcing data centre connectivity makes sense for so many reasons. There’s every chance that the providers you choose (remember you must have more than one if you want true redundancy) will be able to offer you levels of performance, reliability and scalability all but unattainable via the DIY route – and at an extremely attractive price point.

Furthermore, outsourcing your connectivity will give you access to high redundancy, highly regulated, optimised load balancing, guaranteed continuity environments.

Where the do it yourself approach almost invariably leads to a ‘one size bandwidth fits all’ solution, outsourcing provides the very real possibility of matching the bandwidth and latency requirements of each specific application. So, business critical applications that are bandwidth-hungry and require minimal latency, can access a different network than, say, the general office applications traffic. Application-specific bandwidth is a major advantage when it comes to outsourcing.

Availability is another potential major win when it comes to outsourcing versus the in-house option. As outlined earlier, the overall infrastructure required to underpin five 9s availability is expensive to purchase and maintain, and may well be seriously under-used if it is dedicated to one organisation. Access the same infrastructure in an outsourced environment, and you’ll immediately benefit from the economy of scale – the same infrastructure is supporting multiple customers and many more networks, so the cost of accessing it will come down.

Okay, so the Service Level Agreement (SLA) will need to be looked at in detail, but there’s every chance that the outsourcer can provide a level of reliability and resiliency that’s just not affordable and/or practical in-house.

Scaling Up

Scalability is another crucial factor when it comes to comparing your connectivity options. Go it alone and you’re unlikely to have fast, easy access to more bandwidth as and when required. No, you’ll have to take an educated guess at what you want and pay for it, regardless of whether you use it or not. Outsource, and chances are your providers can give you flexible access to more or less bandwidth as required. And, although in its infancy, pay per use (so you only pay for the bandwidth you actually use) is going to offer further flexibility and cost-savings.

Bear in mind that, when it comes to outsourcing connectivity, you do have two main options: carrier neutral data centres and carrier hotels. The carrier neutral option provides you with access to lots of carriers, offering the potential for great levels of flexibility and very competitive pricing, but it does need managing to ensure that you are obtaining the optimum solution at the best price.

In the case of the carrier hotels, you’ll only be offered one ‘flavour’ of connectivity – their own. One contact point and one invoice might compare favourably with the carrier neutral alternative – as does the possibility of true end-to-end connectivity across one organisation’s infrastructure – but the down side tends to be that running a data centre is not the core business of the carrier.

Into the Future

Looking ahead, connectivity to Cloud providers and, in particular, Infrastructure-as-a-Service (IaaS), is set to make a significant impact on the connectivity landscape. Right now, the jury is out as to how the Cloud providers’ connectivity SLAs stack up alongside the colo and carrier data centre ones, but there’s little doubt that the ease of use of the Cloud model will prove a powerful attraction to plenty of organisations already comfortable with the connectivity outsourcing model as outlined earlier in this blog. It’s the next logical step.

Except…I’m becoming more and more convinced that Hybrid ICT is the future. Put simply, this means that, for any enterprise activity that requires some kind of an ICT resource, someone has to carry out an evaluation of the various possible options - taking into account the business criticality of the application alongside the cost, reliability and flexibility of these solutions.

Orchestration (a grander name than ‘management’) is the crucial glue that will bind everything together. And, in the future, you’ll no longer have to decide on the best connectivity solution, you’ll simply be told what to do, based on the application, and the in-house and outsourced options available.